[SciPy-dev] Skypecon Minutes

David Goldsmith d_l_goldsmith@yahoo....
Fri Jul 17 02:05:30 CDT 2009

NumPy Doc Summer '09 Marathon
July 14, 2009 Skypecon Minutes

Start: 18:15 UTC
End:  19:15 UTC

David Goldsmith (Secretary); Ralf Gommers; Joe Harrington; Jack Liddle; Stéfan van der Walt


0)	Status of current effort

Joe expressed general satisfaction with the current state and momentum.  Ralf opined (and no one really argued the point) that much of what is left is either “loose ends,” should be “Unimportant” if it isn’t already, or will require the contributions of persons w/ more in-depth knowledge of NumPy than the average contributor to date.  This motivated the next discussion:

1)	Commencement of work on SciPy doc

Joe expressed the view that we should be well into the review process for NumPy before beginning work on SciPy.  This led to a discussion of the “two review process,” and, mostly, its labor requirements, vis-à-vis required expertise, reviewer vetting criteria and process, etc.; subsequently, a ticket was filed requesting enhancement of the pydocweb infrastructure to support this, and development of some rough guidelines for Reviewers was assigned to David.  Joe also expressed concern over the possibility that SciPy may soon (i.e., on the order of years, since the SciPy doc effort is seen as likely being of the same order) undergo a medium-to-major overhaul, and thus, perhaps, a lot of labor documenting soon-to-be obsolete material may be wasted; Stéfan countered that this shouldn’t be a great concern, because even if there is an overhaul, modules within SciPy are relatively independent of each other and not that much in the way of code will be
 completely discarded, and thus, mostly, if not completely, documenting SciPy as it is presently will not be wasted effort.  Also on the other hand, Jack, with David and Ralf’s support, pointed out that the lack of “pedestrian” docstrings to edit in NumPy, of which SciPy has a plethora, may discourage a lot of the volunteer labor pool, perhaps resulting in another “stall” in activity.  Everyone agreed that this was the greater concern, and it was resolved that work should commence on the SciPy docstring project, beginning with organization of the SciPy material into Milestones ala what we presently have for NumPy, a task which was delegated to Jack.

2)	Discussed the idea of a “User Guide” 

And whether, for instance, SciPy and NumPy should have separate UG’s or be packaged together and designed and organized accordingly.  No resolution.  

3)	Meeting concluded w/ a specific question 

Jack asked for clarification of how to document a function which is basically just a convenience function, in that it thinly wraps another function, merely setting one parameter before “relaying” its call; David noted that this is exactly analogous to the recent discussions about methods equivalent to functions, and the same solution was suggested, namely, document the wrapper function minimally, referring to the wrapped function for details; no opposing view was voiced.

Next Skypecon: Wednesday, July 22, 19:00 UTC


A few “drops,” but relatively minor considering the number of people participating and the technical problems we’ve had in recent Skypecons.


More information about the Scipy-dev mailing list